In the semiconductor industry, void detection in the die attach process is crucial. WeLinkirt provides a locally deployed detection solution for a leading semiconductor manufacturer with its advanced AI vision technology.
User Scenario: A leading semiconductor manufacturer's die attach production line, with high - performance chips as the main products. The detection object is the voids generated during the die attach process of the chips. The existence of these voids can affect the heat dissipation and electrical performance of the chips, thereby influencing the reliability and service life of the chips.
Pain Points: The manufacturer faces many difficulties in void detection. On the one hand, the false - negative rate of traditional detection methods is relatively high, reaching about 3%. As a result, some defective products flow into subsequent processes, increasing production costs. On the other hand, the false - positive rate is as high as 20%, which requires a large number of qualified products to be re - inspected, wasting a lot of manpower and time. In addition, the manufacturer has extremely high requirements for data security and compliance, fearing that the leakage of detection data will lead to the leakage of process secrets. Meanwhile, when changing the production line model, traditional detection methods require a long debugging time, generally about 2 hours, which seriously affects the production efficiency.
Technical Principle
WeLinkirt's DaoAI uses advanced visual basic models and 3D imaging technology to solve the problems. The visual basic model has a powerful feature recognition ability and can quickly learn the features of voids from a small number of samples. Through the APDT positive - sample/few - sample learning method, it can accurately identify voids with only 1 - 20 good samples. This is because the model can perform semantic analysis on the images and extract the key features of voids, such as shape, size, and position, so as to achieve high - precision detection.
- The self - developed 3D camera can obtain the three - dimensional topography information of the chip surface. Through the three - dimensional topography reconstruction technology, the three - dimensional structure of the voids can be clearly presented, effectively detecting hidden solder joints and micron - level topography. This is because 3D imaging can provide more information and can more accurately judge the existence and size of voids compared with traditional 2D imaging.
- The semantic false - positive filtering algorithm combines deep learning and image semantic understanding technology, which can conduct a second - round screening of the detection results. It can identify false positives caused by factors such as image noise and illumination changes, thereby significantly reducing the false - positive rate.
- The advanced algorithm used by WeLinkirt can adapt to different chip models and production environments. When changing the production line model, the system can quickly adjust the parameters to meet the requirements of new products. This is achieved through continuous learning and optimization of the model, enabling the system to automatically adjust the detection strategy according to different product features.
- Local data processing is the key to ensuring data security and compliance. All detection data are processed and stored on the local server without uploading to the cloud, avoiding the risk of data leakage. At the same time, the system uses strict access control and encryption technologies to ensure the security of data.
WeLinkirt's Solution and Product Introduction
WeLinkirt provides the DaoAI AI AOI software system and the DaoAI 2D/3D AI AOI equipment. The DaoAI AI AOI software system has a 0 - code automatic programming function, and programming can be completed in 5 minutes for one good product, greatly shortening the programming time. Its APDT positive - sample/few - sample learning ability can achieve high - precision detection with only a small number of good samples. At the same time, the semantic false - positive filtering function can effectively reduce the false - positive rate. The DaoAI 2D/3D AI AOI equipment uses a self - developed 3D camera and three - dimensional topography reconstruction technology, which can detect hidden solder joints, coplanarity, and micron - level topography, providing more accurate information for void detection. Both products support 100% local private deployment through SDK/API/Docker to ensure that data does not leave the factory, meeting the customer's data compliance requirements.
WeLinkirt's solution not only improves the detection accuracy but also ensures data security, providing reliable support for the semiconductor manufacturer's production.
Quantitative Results: After adopting WeLinkirt's solution, the void detection rate of the manufacturer has increased to over 98%, and the false - negative rate has decreased to <2%. The false - positive rate has decreased by - 70%, greatly reducing the re - inspection volume. The production line model change time has been shortened from the original 2 hours to 5min, significantly improving the production efficiency.